Data-Efficient Quantum Noise Modeling via Machine Learning

Abstract: Maximizing the computational utility of near-term quantum processors requires predictive noise models that inform robust, noise-aware compilation and error mitigation. Conventional models often fail to capture the complex error dynamics of real hardware or require prohibitive characterization overhead. We introduce a data-efficient, machine learning-based framework to construct accurate, parameterized noise models for superconducting quantum processors. Our approach circumvents costly characterization protocols by learning hardware-specific error parameters directly from the measurement data of existing application and benchmark circuits. The generality and robustness of the framework are demonstrated through comprehensive benchmarking across multiple quantum devices and algorithms. Crucially, we show that a model trained exclusively on small-scale circuits accurately predicts the behavior of larger validation circuits. Our data-efficient approach achieves up to a 65% improvement in model fidelity quantified by the Hellinger distance between predicted and experimental circuit output distributions, compared to standard noise models derived from device properties. This work establishes a practical paradigm for noise characterization, providing a crucial tool for developing more effective noise-aware compilation and error-mitigation strategies.